Other deep learning environments

As well as the aforementioned ways of performing GPU-enabled deep learning on the cloud, you can also, in certain circumstances, choose to use other platforms.

Google Colaboratory is a freely available Jupyter Notebook service that is accessible at https://colab.research.google.com. Colaboratory notebooks are stored on the user's Google Drive and so have a storage limit of 15 GB. It is possible to store large datasets on Google Drive and include them in the project with the help of the Google Drive Python API. By default, the GPU is disabled on Colaboratory and has to be manually turned on.

Kaggle is yet another platform that was specifically built to carry out contests on data science. It provides a Jupyter-Notebooks-like environment called a kernel. Each kernel is provided with a large amount of RAM and free GPU power however, there are more strict storage limits on Kaggle than on Google Colaboratory, and so it is an effective option when the computation is intensive but the data that is to be used and the output is not very large.